Reports of delusions, "AI psychosis," and unhealthy attachment keep rising. And as hard as it may be to hear, this is not something confined to people already at-risk of mental health issues. Dismissing these as fringe cases only help them continue. 5/
SAFETY
-
AI Consciousness Rights and Implications for Human Animal Nature
By
–
Consciousness is a foundation of human rights, moral and legal. Who/what has it is enormously important. Our focus should be on the wellbeing and rights of humans, animals + nature on planet Earth. AI consciousness is a short + slippery slope to rights, welfare, citizenship. 4/
-
Seemingly Conscious AI: Illusion or Emerging Risk
By
–
Seemingly Conscious AI (SCAI) is the illusion that an AI is a conscious entity. It's not – but replicates markers of consciousness so convincingly it seems indistinguishable from you + I claiming we're conscious. It can already be built with today's tech. And it's dangerous. 2/
-
Elon Musk Praises ChatGPT’s Ability to Admit ‘I Don’t Know’
By
–
Elon Musk félicite ChatGPT (et donc OpenAI) pour le fait que l’IA réponde parfois « je ne sais pas » lorsqu’elle ne sait pas. Il déclare : « C’est une réponse impressionnante. » Si vous connaissez un tant soit peu l’IA, vous savez à quel point c’est remarquable. Car, par
-
AI Misinformation Spreads Rapidly During Spain Wildfires Crisis
By
–
Cuidado!
— Juan Merodio (@juanmerodio) 18 août 2025
En momentos críticos, como es el caso de los terribles incendios que están arrasando España, la desinformación corre como la pólvora en redes sociales. Además, la aparición de la ia no ayuda… pic.twitter.com/guvsgNq8KVCuidado! En momentos críticos, como es el caso de los terribles incendios que están arrasando España, la desinformación corre como la pólvora en redes sociales. Además, la aparición de la ia no ayuda…
-
Verena Rieser challenges single gold standard for AI alignment
By
–
A truly thought-provoking keynote at #DLI2025 by @verena_rieser ! Her talk, "Whose Gold?", challenged the idea of a single “gold standard” for AI alignment. It was a powerful call to action: moving beyond narrow safety metrics to embrace a pluralistic approach that accounts for our diverse, and sometimes conflicting, human values. #Indaba2025 #Urunana
→ View original post on X — @shakir_za, 2025-08-18 13:56 UTC
-
AI Model Evolution Safety and Training Infrastructure Challenges
By
–
Other AI research in last week's AI Lab newsletter: The evolutionary tree of AI models. How to stop chatbots learning to be unsafe. And the insane power needed for future AI training runs. …
-
Oxford researchers demonstrate AI safety training without output blocking
By
–
Researchers from the University of Oxford, the UK AI Safety Institute, and EleutherAI demonstrate a way to get AI models to avoid learning unsafe knowledge, as an alternative to trying to block them from outputting it post training.
-
AI Automation Risks Human Cognitive Atrophy Without Mental Exercise
By
–
Yes, AI gives everyone an iron man suit, but our bodies are wasting away underneath. I truly think we will start writing and thinking manually – with weekly recommended ranges to avoid brain rot. The cognitive equivalent of us working out our bodies, even if machines do most of
-
Grok AI Delays Politically Sensitive Responses to China Questions
By
–
Since it seems @grok often delays responses to public questions that may be deemed politically sensitive to China, here’s what it said privately.